103
Resolution and the size of the data set likewise increase as sampling frequency
of the ADC increases. Converter sampling rate is commonly quoted as samples-persecond (sps). In order to capture the variability of the signal, the sampling frequency
must be at least half the period of the highest frequency to be sampled. Current
technology allows sampling rates up to 300 Msps allowing reliable sampling of
frequencies up to about 150 MHz, precision rarely required for environmental
applications. In summary, as resolution in time and in sensor response readout
increases, precision increases but the data set becomes larger. This tradeoff has
prompted a continuing race between data generation rate and data storage and handling capacity. Advances in circuit miniaturization and technology development
have fortunately until now relieved this bottleneck.
An effective environmental sensor is one capable of measuring a signal across its
natural analog span at numerical resolution commensurate with the application at
hand. The time-referenced analog output of an effective sensor will result to be
graphically representable as a smooth succession of peaks and valleys of different
frequencies and amplitudes. In practice, single digit resolution (0–9) is considered
poor, so sensor output values are usually quoted with a precision of two or three decimal digits and only rarely (and with sufficient justification) at greater resolution.
Fourier theory proposes that any such time series can be represented as a composite of superimposed sinusoidal variations. Breaking down a signal into component
frequencies throughout this spectrum allows a simple solution to ameliorate signal
noise; the application of frequency cutoff filters. Rapid positive and negative fluctuations superimposed upon a long smooth wave representing a low frequency signal
can be removed by cutting off variations at frequencies above that low frequency
band of interest, a procedure known as the application of a low band pass filter. A
high frequency signal superimposed upon low frequency noise may be conditioned
by filtering out only low frequency variations constituting a high band pass filter.
Cutoff band pass filters can isolate the band sought or alternately blank out a noisy
band in an otherwise informative spectrum. Early electronic filters used circuits composed of capacitors, induction coils, and resistors to dampen out the unwanted signal.
Various piezoelectric microelectromechanical devices are in use today.
Electronic Noise
Noisy data may require electronic filtering. Use of the term noise arises of
course from the field of audio reproduction and the analog audiophile recognizes electronic noise as the hiss of the needle or the tape, the scratches on the
vinyl, or the gaps on the tape. When the noise surpasses the signal, the recording can no longer be distinguished. Since the signal-to-noise ratio dictates the
resolution, very noisy signals must be filtered prior to quantization.
The Global Telecommunications System (GTS) is a major component of transmitting global meteorological data, consisting of both in situ and satellite observations. This data is collected by a number of organizations which archive and further
process the data. In the US, the National Centers for Environmental Prediction
5.1 Data Signal Conditioning for Ocean Observing
Resolution and the size of the data set likewise increase as sampling frequency
of the ADC increases. Converter sampling rate is commonly quoted as samples-persecond (sps). In order to capture the variability of the signal, the sampling frequency
must be at least half the period of the highest frequency to be sampled. Current
technology allows sampling rates up to 300 Msps allowing reliable sampling of
frequencies up to about 150 MHz, precision rarely required for environmental
applications. In summary, as resolution in time and in sensor response readout
increases, precision increases but the data set becomes larger. This tradeoff has
prompted a continuing race between data generation rate and data storage and handling capacity. Advances in circuit miniaturization and technology development
have fortunately until now relieved this bottleneck.
An effective environmental sensor is one capable of measuring a signal across its
natural analog span at numerical resolution commensurate with the application at
hand. The time-referenced analog output of an effective sensor will result to be
graphically representable as a smooth succession of peaks and valleys of different
frequencies and amplitudes. In practice, single digit resolution (0–9) is considered
poor, so sensor output values are usually quoted with a precision of two or three decimal digits and only rarely (and with sufficient justification) at greater resolution.
Fourier theory proposes that any such time series can be represented as a composite of superimposed sinusoidal variations. Breaking down a signal into component
frequencies throughout this spectrum allows a simple solution to ameliorate signal
noise; the application of frequency cutoff filters. Rapid positive and negative fluctuations superimposed upon a long smooth wave representing a low frequency signal
can be removed by cutting off variations at frequencies above that low frequency
band of interest, a procedure known as the application of a low band pass filter. A
high frequency signal superimposed upon low frequency noise may be conditioned
by filtering out only low frequency variations constituting a high band pass filter.
Cutoff band pass filters can isolate the band sought or alternately blank out a noisy
band in an otherwise informative spectrum. Early electronic filters used circuits composed of capacitors, induction coils, and resistors to dampen out the unwanted signal.
Various piezoelectric microelectromechanical devices are in use today.
Electronic Noise
Noisy data may require electronic filtering. Use of the term noise arises of
course from the field of audio reproduction and the analog audiophile recognizes electronic noise as the hiss of the needle or the tape, the scratches on the
vinyl, or the gaps on the tape. When the noise surpasses the signal, the recording can no longer be distinguished. Since the signal-to-noise ratio dictates the
resolution, very noisy signals must be filtered prior to quantization.
The Global Telecommunications System (GTS) is a major component of transmitting global meteorological data, consisting of both in situ and satellite observations. This data is collected by a number of organizations which archive and further
process the data. In the US, the National Centers for Environmental Prediction
5.1 Data Signal Conditioning for Ocean Observing